50,000 HR LED Life

So many LED’s are rated for 50,000 hours. This is about 2,100 days. My questions is , is it like MTBF for hard drives? Is it regardless of power level? Does it assume a certain duty cycle, for example 30 min per days? Is the rating really a few hundred hours with the possibility of one or two in the lot reaching 50,000? Does it mean if you samples 100 of the same LED, half would last more than 50,000 hours, half would last for less?

I get 10 year life on outdoor led light running 5 hrs per day. Everyday 24x7.

For those LED ratings, they assume the led is run within spec (so spec’d amps and 85°C or below). They are then claiming the output will be some % of rated after that many hours (I think somewhere 70-75-80% range typically). Continuous or not, the LED isn’t going to care much.

Ok, so the lumens get reduced as you use it. Interesting

Yeah, it’s not a failure rating. It’s 50,000 hours to 70% original output. The durability is amazing.

A 30% lumen depreciating is losing nearly 1/3 of the light. That’s not good. That could easily be a lighting install meeting spec. to being under-lit, and no-one really notices how much darker the room is. T8 fluorescent had a 90-95% lumen maintenance. Incandescent was also fairly high.

Except when I was working in commercial lighting, the specs were accounted for (and mentioned) when quoting installations.

It goes far beyond that. The commercial LED industry has long used Lumen Maintenance Standards in certifying lumen degradation curves for emitters. I ran across this while shopping for +$1200/unit LED high-bay lighting and became concerned that the mainstream commercial manufacturers never spec’d Cree emitters. A bit of research showed lumen maintenance comparisons and how comparable Cree emitters dimmed more quickly than their competitors (and failed lumen maintenance certification standards). While the typical home owner doesn’t fuss over such certifications while shopping for LED lighting, addressing this spec becomes critical in the commercial environment, where labor cost and potential downtime become a huge factor in maintaining a lighting infrastructure. This is why the best commercially proven light housings utilize a vast number of emitters driven at or below manufacturer specs, with massive efficient heat sink designs to keep the emitters at or blow their Lumen Maintenance Certification specs, with expensive constant current drivers that provide low current/voltage ripple to supply clean stable power throughout the life of the emitters… ie: not the Chinese junk sold online or through brick and mortar stores.

I recently purchased several 35,000 lumen RAB area spot lights that utilizes a passive heat sink design that create its own thermal updraft to keep air moving through the heat sink without a fan. The science behind properly designed LED light housings is impressive, with manufacturers spending large amounts of capitol on research centers to create and prove their designs.

Flashpilot - very interesting info. Thanks for sharing

I can imagine a light like the FW3A running hot on turbo would only last in the 5000 hr range.

Except that in doing this you wind up with installations that are substantially over lit in the beginning, much more so then T8 fluorescent. This wastes more power over time. You worked in commercial lighting, you should know LED has nearly the same lumen maintenance of metal halide a light source with lumen maintenance even worse then T5, but LED goes on for longer in its deeply depreciated state. Furthermore, IES advises against over-lighting in street light installs to accommodate for light loss in its newest RP8. What you should do is run something like a CLO driver, constant lumen output. It’s a technology that gradually increases the power to the fixture as it hits certain age milestones to accommodate for loss.

Keep in mind human brightness perception is non-linear. Two bulbs next to each other at 100% and 70% brightness will be clearly distinguishable as different, but in an isolated context, the difference is relatively subtle.

A decline of 30% output over the life of an installation is not going to make or break the utility of a space. L70 life ratings are also not an absolute criteria. They are a uniform basis for comparison. If a lighting designer needs to ensure maximum uniformity of lighting, they’re going to pick lights with a long L70 rating and they’re are going to specify replacement long before they reach that life. Or these days they probably have a reasonable number of options for CLO drivers like you suggest. Everyone else is going to just accept the gradual decline in output.

Fluorescent lamps comparisons have to be made conditionally, because the most common end-of-life criteria is usually outright failure, although I’ve also seen cheap CFL’s dim unacceptably over a few years. For example, here’s a datasheet for a Philips T5 lamp, and as a bonus, Philips gave lumen maintenance factors.

They show 90% lumen maintenance at 20,000 hours, with 84% of lamps remaining functional at that time. They also project a rapid decrease in survival factor shortly afterwards as the lamps approach wear out - 50% functional at 24,000 hours.

Good commercial LED luminaires are typically rated 50,000 hours for L70 life, and I’ve seen as high as 100,000 hours. I don’t know if there is a formal criteria for lamp survival factor, but in some contexts I have seen 90% survival specified.

Practically speaking, lights seldom get replaced when they’re dim. They get replaced when they die.

You’re welcome.